AI & intelligence
Claude AI integration
How does AonHive use Anthropic Claude?
Claude AI integration for reasoning and knowledge workflows with explicit controls.
Claude is a model we put on a defined task: read, draft, or classify. The prompt is versioned against a set of cases for that task. A change that fails the set does not ship. Sending money, deleting a record, or writing to a customer waits for a person.Discuss this stackHow we use it
A task, a set, and a person on the risky branch.
- 01
Keep the case set next to the prompt
Routine inputs, missing context, and the case that must refuse are in the set. A single good demo is not the evaluation.
- 02
Scope the tools
A tool reads or writes one kind of record. The model does not receive a credential that can do more than that tool.
- 03
Show the failure
A timeout or a low-confidence result is a state the user can see. The workflow does not fill the gap with a confident guess.
Anthropic Claude
Bring the system this has to fit.
Tell us the product, the estate you already run, and what your team needs to own. We will say whether this belongs in the stack.